- Add to variables only once, instead of from within a loop.
- Use :tl instead of :M to match against strings for readability.
- Use CLEANFILES instead of a custom clean target.
Full release built successfully after this change.
Upstream sources can be fetched by running "make checkout" in
src/external/bsd/llvm, they will be properly imported once the
integration and missing features are sorted out.
NetBSD/emips port runs on Xilinx and Beecube FPGA systems and the
Giano system simulator.
eMIPS is a platform developed at Microsoft Research for researching
reconfigurable computing. eMIPS allows dynamic loading and scheduling
of application-specific circuits for the purpose of accelerating
computations based on the current workload.
NetBSD eMIPS support for NetBSD 4.x was written at Microsoft Research
by Alessandro Forin and Neil Pittman. Microsoft Corporation has
donated full copyright to The NetBSD Foundation.
Platform support for eMIPS is the first part of Microsoft's
contribution. The second part includes the hardware accelerator
framework and will be proposed on tech-kern soon.
Honour this for dependency processing in bsd.dep.mk. Switch i386 and
amd64 assembly to use ISO C90 preprocessor concat and drop the
-traditional-cpp on this platform.
the debug symbols and adding the debug-link to .debug.
Use '(rm -f file; false)' in the failure path to force failure.
Based on solution proposed by Nicolas Joly on tech-toolchain in July 2010.
Should fix PR toolchain/44046 from Andreas Gustafsson.
tre will be compiled without approx and wchar/mulibyte support to
only match the minimum requirement to replace our spencer regex.
This needs a lot of testing.
Only enabled when USE_LIBTRE is set to `yes'.
- Designed to be fully MP-safe and highly efficient.
- Tables/IP sets (hash or red-black tree) for high performance lookups.
- Stateful filtering and Network Address Port Translation (NAPT).
Framework for application level gateways (ALGs).
- Packet inspection engine called n-code processor - inspired by BPF -
supporting generic RISC-like and specific CISC-like instructions for
common patterns (e.g. IPv4 address matching). See npf_ncode(9) manual.
- Convenient userland utility npfctl(8) with npf.conf(8).
NOTE: This is not yet a fully capable alternative to PF or IPFilter.
Further work (support for binat/rdr, return-rst/return-icmp, common ALGs,
state saving/restoring, logging, etc) is in progress.
Thanks a lot to Matt Thomas for various useful comments and code review.
Aye by: board@
This change adds the ATFFILE variable, which can take the following values:
- yes: The source directory contains a static Atffile to be installed.
- auto: The Atffile is automatically generated.
- no: Do not install any Atffile (replaces the previous NOATFFILE variable).
The ugly bit of this is that I have to add a TESTS_SUBDIRS variable that
lists the subdirectories that contain test cases (so that they can be listed
in the Atffiles, without including auxiliary directories without tests).
This is a subset of SUBDIR.
Addresses a minor part of PR bin/43394.
we're ELF now, and there are many missing checks against OBJECT_FMT.
if we ever consider switching, the we can figure out what new ones
we need but for now it's just clutter.
this doesn't remove any of the support for exec_aout or any actually
required-for-boot a.out support, only the ability to build a netbsd
release in a.out format. ie, most of this code has been dead for
over a decade.
i've tested builds on vax, amd64, i386, mac68k, macppc, sparc, atari,
amiga, shark, cats, dreamcast, landisk, mmeye and x68k. this covers
the 5 MACHINE_ARCH's affected, and all the other arch code touched.
it also includes some actual run-time testing of sparc, i386 and
shark, and i performed binary comparison upon amiga and x68k as well.
some minor details relevant:
- move shlib.[ch] from ld.aout_so into ldconfig proper, and cut them
down to only the parts ldconfig needs
- remove various unused source files
- switch amiga bootblocks to using elf2bb.h instead of aout2bb.h
building X11 without inet6 support. the support to maintain patches to
generated files and other files is too great. PR#42856
- add TOOL_PIGZ, which defaults to the (not yet default) built tools/pigz.
- kill MKPIGZ.
- add MKPIGZGZIP. if it is "no", then only install pigz as pigz. if it is
not "no", don't install mrgzip, install pigz as gzip.
in all cases, the z* scripts.
i don't like the name MKPIGZGZIP so if anyone has a better name, feel free
to replace it. i don't care enough since killing the ugly name is just
more incentive to get pigz doing .bz2, .Z and .z.
make test may yield misleading results but should work in many cases. This
target is supposed to simplify the execution of tests but does not preclude
developers from running the installed versions.
Addresses PR misc/38326.
the diff -u is more obvious than a description:
-X11FLAGS.VERSION= -DOSMAJORVERSION=1 -DOSMINORVERSION=6 # XXX
+X11FLAGS.VERSION= -DOSMAJORVERSION=5 -DOSMINORVERSION=99 # XXX
man pages to use mandoc unconditional as it gives reasonable output for
all man pages, not only a subset of mdoc(7). Use the newly installed
style.css for formatting and produce hyperlinks for .Xr.
While this is not strictly sensible, it keeps UPDATE=1 builds
rolling if I change some aspect of config (such as adding the
cfdriver/cfattach lists now). This is going away eventually.
rump into share/mk. This is to make it useful for all kernel
builders.
Note: we have waaay too many weird and wonderful ways of making
kernel code (monolithic kernel, modules, rump). There should be
only one way to build kernel code instead of a maze of twisty little
.mk files, all not quite alike. When that is fixed, this snippet can go
into the more generic .mk file.
conflicts with shared libaries names libXX.so; as recently seen with
MKUPDATE=yes builds for libelf and libnvpair.
All now stalled regular .so files need to be manually removed from
object directories.
relocations by the module loader in the kernel. This makes modules work
in, at least, shark and macppc respectively.
This is obviously a workaround that results in slower code, but at least
makes modules work. I'm adding a comment detailing what the real solution
would be so that the whole thing can be revisited in the future. (Read:
I don't have time now to dig the details of how to implement trampoline
generation.)
(Based on the old bsd.kmod.mk file, I understand that hppa needs this
workaround too, but I can't check this platform.)
decide whether to make kernel modules: set MKKMOD to no for evbppc.
Use this in etc/Makefile to decide whether to do the "modules"
obsolete sets. Move the ./var/db/obsolete/modules entry from the
"mi" to the "module.mi" file set.
Fixes the build for evbppc.
Discussed with uebayasi@
which potentially have multiple output files, which confuses make(1).
This wrapper copies all involved files into /tmp, execute the command, and
copy the wanted file back to ${.OBJDIR}. If this is expensive, consider to
rewrite the generation command to output files individually.
XXX bsd.own.mk is too bloated.
to ${LD} via:
${LDADD.${PROG}}
${LDFLAGS.${PROG}}
${LDSTATIC.${PROG}}
${LDADD.${LIB}}
${LDFLAGS.${LIB}}
OTOH you can't pass parameters to ${CC}, because in suffix rules make(1) only
knows the name of ${.IMPSRC} and ${.TARGET}; it's users' responsivility to
define ${CC} parameters to all the sources of a given ${PROG} / ${LIB}.
Should address bin/42381.
(Bug in this commit log was pointed out by mrg@.)
default of this platform. This fixed the build for NetBSD/amd64
and NetBSD/i386.
Yes, this horrible. There should be a better way to handle these
settings in general.
that also have the wsmouse driver.
xf86-input-ws adds support for touchpanels in X11 that report absolute position
data through wscons(4), e.g. the extended ums(4) driver. xf86-input-ws was
written by Matthieu Herrb for OpenBSD.
Please remember to also update xsrc before the next system build.
don't add ${CPPFLAGS_ISYSTEM} or ${CPPFLAGS_ISYSTEMXX} to refer
${DESTDIR}/usr/include or ${DESTDIR}/usr/include/g++.
This change might cause errors on some MD stand dir, but in that case
each Makefile should be fixed to search proper system include paths
by -Ipath option in own CPPFLAGS.
SSP for building the whole source tree. Set this to "yes" for
NetBSD/amd64 and NetBSD/i386. It is now possible to get the old
behavior (build libs and certain programs with SSP) by setting
"USE_SSP_DEFAULT" to "no".
NetBSD/i386 as previously discussed on the "port-amd64" and
"port-i386" mailing lists. No objections from the core team.
Thanks a lot to Tobias Nygren for helping me with the benchmarking
and to Christos Zoulas for fixing SSP problems in cgd(4).
makes {MK,HAVE_}BINUTILS consistent with {MK,HAVE_}{GCC,GDB}.
Allow MKBFD to defines MKBINUTILS as a backwards compatibility hook.
Update the sets lists and add conditionals for lib{bfd,opcodes}.
update the pkgconfig sed script to handle newer pixman.
XXX: should make the pkgconfig sed script barf it it ends up leaving
XXX: behind something that looks like "@.*@", i think.
when installing hard links. They have no effect except when using a
metalog, in which case the information is added to the metalog. In
the future, these variables may be replaced by a method for explicitly
recording hard links in a metadata log.
Also change a few things that called ${INSTALL_LINK} without going
through bsd.links.mk.
Reviewed by perry and joerg. This should fix PR 24457 and PR 41155.
can use instructions which were not available on the original i386
(eg cmpxchg). Due to some strangeness in gcc's i386 support this needs
an extra --with-arch=i486 configure argument for gcc to have the desired
effect, see my post "i386 vs i486, some inconsistencies" to tech-toolchain
some weeks ago.
I'm not happy to break compatibility, but since (a) kernel support
for i386 was removed and (b) i387 code was put into libm this is
just another coffin nail.
The gain is besides consistency and more efficient code that intel
atomar intrinsics can now be used by gcc. (which would need runtime
library support otherwise)
is set to "yes" -- defaults to "no" except for build.sh builds. This
results in a deterministic .a file rather than one that reflects
timestamps and permissions on the source files.
Also, clean up the ar flags we're using, and remove a redundant use of
ranlib that on a modern POSIX ar can be done with the "s" flag.
Discussed on tech-toolchain
prints something that does not look like an identifier, then use uname
-m instead. (Cygwin prints "unknown", and OpenBSD prints a long string
containing several spaces; this code should handle both.)
Test compiled on i386 and amd64; there may be some stragglers
on other platforms.
Note: -Wall has this by default in gcc4, and we explicitly disable it at
WARNS=1 (i.e., -Wall -Wno-sign-compare). A goal is to reduce the WARNS level
where this feature is enabled, so we can eventually remove -Wno-sign-compare.
(ev56, handled as non-BWX by Xorg) and XP1000 (ev6, BWX). Also
tested by Rafael Ruiz on a AlphaStation 255 (thanks !).
non-PCI adapters are not supported; someone should make these work
as a wsfb device.
library. This is mostly a convenience, so that you can trigger
a shared library rebuild by touching the shlib_version file, it
should not otherwise impact the build one way or the other.
shared library. This is done so that -L options pointing into
DESTDIR will come after -L options pointing into our object tree
for shared libraries this shared library depends on.
This makes a difference when shared library major numbers are bumped
(as was recently done in our tree), and you build into an already-
populated DESTDIR, because otherwise the old major version shared
libraries will be picked up, because the new ones have not yet been
installed at this stage. This will in all probability lead to
conflicts later on when linking programs, where one would try to
mix new and old major versions for the same shared library.
I *hope* this will not have any negatively impact by moving other
order-dependent options around; local tests with rebuilds did not
uncover any problems I could see.
OK'ed by lukem@
the base NetBSD system. It uses Linux LVM2 tools and our BSD licensed
device-mapper driver.
The device-mapper driver can be used to create virtual block devices which
maps virtual blocks to real with target mapping called target. Currently
these targets are available a linear, zero, error and a snapshot (this is
work in progress and doesn't work yet).
The lvm2tools adds lvm and dmsetup binary to based system, where the lvm
tool is used to manage and administer whole LVM and the dmestup is used to
communicate iwith device-mapper kernel driver. With these tools also
a libdevmapper library is instaled to the base system.
Building of tools and driver is currently disable and can be enabled with
MKLVM=yes in mk.conf. I will add sets lists and rc.d script soon.
Oked by agc@ and cube@.
0 Minimal output ("quiet")
1 Describe what is occurring
2 Describe what is occurring and echo the actual command
3 Ignore the effect of the "@" prefix in make commands
4 Trace shell commands using the shell's -x flag
The default remains MAKEVERBOSE=2.
- introduce X11FLAVOUR to choose src/x11 vs src/external/mit/xorg
for the X11 to build if MKX11=yes is set. it takes the values
of either Xorg or XFree86.
- default to Xorg on alpha, i386, macppc, shark, sparc and sparc64
- remove MKXORG_WITH_XSRC_XSERVER, unused and never useful
Previously, they were defined in bsd.own.mk if USETOOLS=yes, but in
bsd.sys.mk if USETOOLS!=yes. This caused makefiles that did this:
.include <bsd.own.mk>
FOO != ${TOOL_BAR} args...
to work in the USETOOLS=yes case but not in the USETOOLS!=yes case.
set to /usr/bin/bash if HOST_CYGWIN was defined, but now build.sh
tries to set HOST_SH appropriately.
Remove the HOST_CYGWIN variable, which was not used for any other purpose.
Document that HOST_SH should be an absolute path.
THis was proposed in tech-toolchain.
- .man.pre files are treated just like .man files.
- Makefiles are offered the choice of using sed or cpp for the
transformation. MKXORG will default to sed, MKX11 to cpp.
- At least for now, versions of packages are not tracked individually.
